Delos Data, a startup funded by former Intel and Barefoot Networks executives, demonstrates a modular server platform designed to help AI chip startups reach rack scale. The platform includes 36 OSFP ports and OAM sockets, offering a hardware configuration that the company presents as a shortcut to rack-scale deployment.
Delos Data announces a new cluster architecture and server design aimed at accelerating AI inference through cabled scale-up. The design pairs its Nonstop AI approach with Delos Mosaic software, which provides visibility into scale-up domains.
Delos Data unveils a modular server platform for AI chip startups. It enables flexible scale-up domains that can be tailored to AI inference workloads at scale.
Delos Data announced a cluster architecture and server design intended to accelerate AI inference by using cabled scale-up. The design aims to reduce GPU idle time caused by data center interconnect bottlenecks and scale to thousands of GPUs, CPUs, and accelerators.
Delos Data aims to enable AI clusters with more than 1,000 GPUs in a single scale-up domain through its Nonstop AI cluster-management software and modular server design. The company targets lower cost and power per token for AI inference.